The Sony NEX FS700 in post production for VS was relatively straight forward, however as the NEX-FS700 is a new camera and many are interested in the merits of the 4K sensor and new cine curves, I thought it worthwhile to talk about our post workflow for this project.
We shot VS on a pre production NEX-FS700 with Carl Zeiss CP.2 lenses. We were told by Sony that the image wasn't finalised yet and the engineers were still working on finessing the quality out of the camera. That said the increased pixel count on the new sensor married with the CP.2s resulted in a very sharp and detailed image.
The recording format was both internal using the native AVCHD codec at 24Mbs and also an external recorder the Atmos Samurai was used via HD-SDI. The additional HD-SDI output on the camera is extremely helpful and saves constantly having to split a single output to multiple devices. This way we fed the HDMI to a SmallHD DP6 with the HD-SDI going to the Samurai and looping through to a Sony 17" OLED.
The 24Mbs AVCHD codec is much more efficient than expected and produces a very clean output for a majority of situations. In fact often when A/B'd against a higher bit rate codec, it is almost impossible to tell the difference. However, extremes in both lighting and movement can cause the native AVCHD codec to break up and artefact. In the case of live concert recording I would always use a secondary external recording device, both as backup but also to employ a higher bit-rate recording to reduce these artefacts. I only really noticed one shot which suffered image break up with the AVCHD codec, so for these specific frames, only 7 frames in total, I used the ProRes files from the Samurai.
The reason for not using the ProRes media primarily within the edit were two fold; primarily the signal recorded to the Samurai was interlaced as a result of recording at both 50fps in slow and quick mode and 200fps in super slow mode. The Samurai simply won't record a progressive signal at these frame rates. We also discovered that as the camera was still only a pre production model, the output, even when set to clean, would still record camera data, specifically any time I hit the expand focus button. We're hopeful this will be fixed at the time of release and fed this information back to the engineers.
The edit was done in Final Cut Pro 7 with the AVCHD media ingested via Log and Transfer to ProRes 422. Once the edit was locked I transferred an XML file into DaVinci Resolve for the grade.
I had pre graded some shots within FCP during the offline cut to get a feel for the grade I wanted to see. This information isn't carried into Resolve but is helpful in creating a starting point come the grade. I'll often also photoshop up reference grade looks for the client to help in getting an agreed look on projects. Inside of Resolve I can reference both the offline cut and also any other images. I use a library of looks and images for reference which I've built up over different jobs.
My aim of the grade and final look of the film was that of a music video/commercial finish, something stylised and bold in direction. I always start by grading a chosen master image which defines the look and emotion of the film, in this case the master mid shots of Sara the dancer in full costume.
The nature of the timing on the shoot meant that these mid and close up shots were filmed 8 hours into the shoot and my main aim in the grade was to retouch all of Sara's skin to both soften it and remove sweat. The flexible nature of Resolve's node based interface meant I could use both composite overlay modes to blur certain parts of the image as well as keying skin tones for selective softening. I use Resolve primarily due to this flexibility over the image. I also created individual shapes to isolate corrections over the eyes bringing out brightness on some shots and tracking these shapes to the movement of the image.
I enjoy the power inside Resolve of being able to heavily alter and selectively grade different parts of the image, however it is also essential to be able to work through the whole timeline without getting stuck for hours on one or two shots. We did this project on our own time with no one paying per hour, however time was still limited and I had around 6 hours allocated to doing the grade. Once the master shots had been graded, it was then a process of working through the rest of the timeline, correcting and selectively working on parts of the image to create a consistent look and feel to the film.
I am constantly impressed just how well the native ACVHD codec of both the FS100 and also the FS700 has been in post. There are many times when recording to an external recorder just isn't possible and in these cases you are completely reliant on what you can pull from the 24Mbs codec. I believe that the new cine gammas within the FS700 also help with increasing dynamic range within the camera as there have been a couple of shots in other FS700 footage I've graded which the highlights actually contained more information than I believed.
There are of course other cameras on the market which will do higher frame rates and have better dynamic range however, none of them come close to the price of the FS700. For this reason alone, I plan to be using as FS700 as soon as it becomes available and look forward to watching the developments and upgrade to a full 4K signal out of the camera in the future.